Synergistic Effects of Calcium Chloride and Polypropylene Fibre on Khanote Soil: Experimental and AI Modelling

Abstract

This study investigates the stabilization of marginal Khanote soil (A-1) using Calcium Chloride and Polypropylene Fibre, coupled with Artificial Neural Network predictive modelling. Laboratory evaluations involved Modified Proctor and Unconfined Compressive Strength tests across varying additive dosages. Results demonstrate that 1% CaCl₂ significantly improves compaction, increasing maximum dry density from 1960 kg/m³ to 2130 kg/m³ through hygroscopic densification. Conversely, polypropylene fibre acts as a mechanical bridge, with an optimal 0.5% dosage yielding a peak UCS of 4.12 MPa, effectively transitioning the soil from brittle to ductile failure. A synergistic mix of 1% CaCl₂ and 0.5% polypropylene fibre produced a robust UCS of 3.95 MPa, balancing chemical density with mechanical elasticity. To optimize mix designs, a multi-layer perception ANN is developed, achieving an overall correlation coefficient (R²) of 0.988 and low error metrics (RMSE = 0.079 MPa). The AI model successfully captured non-linear behaviours, including strength degradation associated with fibre agglomeration at 1.5% fibre content. This research provides a high-precision, data-driven framework for soil stabilization, offering civil engineers a sustainable tool for infrastructural development in arid regions by minimizing trial-and-error laboratory cycles while ensuring superior structural integrity. The findings validate AI as a transformative asset for modern geotechnical engineering.